Title: Customised U-net model-based brain tumour segmentation in MRI images and ensemble-based tumour classification systems

Authors: P. Devisivasankari; K. Lavanya

Addresses: School of Computer Science and Engineering, Vellore Institute of Technology, Vellore, 632014, Tamil Nadu, India ' School of Computer Science and Engineering, Vellore Institute of Technology, Vellore, 632014, Tamil Nadu, India

Abstract: Medical image processing requires autonomous brain tumour segmentation because early diagnosis can improve survival by treating brain cancers quickly. Brain tumours are manually classified by experts, which is time-consuming. Brain tumour (BT) diagnosis takes time and skill, hence radiologists must be skilled. As patient numbers have grown, so has data volume, making outdated methods expensive and inefficient. Many scholars have studied fast and accurate BT detection and classification algorithms. DL can locate BTs in medical photos using trained convolutional neural network (CNN) models. Brain tumour segmentation is easier with automatic segmentation, which is widespread. This work categorises and automates brain tumour segmentation using customised UNet model-based brain tumour segmentation (CUNet-BTS). Classification, preprocessing, segmentation, feature extraction, and fusion are modelled. Gaussian filtering enhances MRI pictures. Finally, an ensemble classification algorithm is suggested. For classification, this model combines the output scores of optimal DeepMaxout, DCNN, and RNN classifiers. The excellent training model Pelican Assisted Chimp Optimisation (PACO) Method can change classification model weights.

Keywords: MRI; magnetic resonance imaging; CNN; convolutional neural network; intersection-over-union; FCNNs; fully convolutional neural networks; recurrent regression based neural network; internet of medical things.

DOI: 10.1504/IJSSE.2025.151320

International Journal of System of Systems Engineering, 2025 Vol.15 No.6, pp.529 - 560

Received: 27 Jun 2023
Accepted: 04 Sep 2023
Published online: 23 Jan 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article